Objective To examine game-related factors that are most reliably associated with concussions in which players do not demonstrate on-ice visible signs.

Design Retrospective cohort.

Setting National Hockey League (NHL) Regular Season Games over two seasons (2014–15 & 2015–16).

Participants NHL players diagnosed with concussion; analysis was limited to events where digital video of the injury was available (n=151).

Interventions (or Assessment of Risk Factors) Game-related factors: ice location, time zone change from prior game, time of season, hits received and given within the same game and the last 7 days, and whether the injured player played the previous day.

Outcome Measures Both univariate and multivariate analyses were completed to examine differences in game-related factors between concussed players with no visible signs (-VS) and those with visible signs (+VS).

Main Results Univariate analysis between the -VS and +VS groups found no statistically significant differences. PLS discriminant analysis found that concussions with -VS were associated with fewer hits given in the last 7 days (bootstrap ratio = 2.3, p = 0.02), fewer hits given in the last game (bootstrap ratio = 2.3, p = 0.02), and fewer total hits in the last game (bootstrap ratio = 2.1, p = 0.03).

Conclusions The present study found that players who sustained a concussion with no visible signs had fewer hits given and fewer total hits in the proceeding game(s).